Calculating the Project Buffer

The Critical Chain Project Management method recommends to make uncertainty explicit using buffers to protect critical and nearly critical paths in the project network schedule diagram. Expliciting buffers is a quite interesting practice for project governance, since KPIs like percent completed buffer can be easily monitored graphically, like this chart shows:

In this post we will see how to calculate the project buffer, following an example proposed by Mike Cohn: Agile Estimating and Planning, Prentice Hall, 2012.

Imagine you need to calculate how long before you have to leave home to drive to the airport. If you estimate these five activities, with their durations and local safeties (that is, 90% estimate minus 50% estimate) then you should leave 2 hours and 50 minutes before boarding time:

If you keep just a confidence level of 50% for each activity, then you should leave only 1 hour and 10 minutes before:

Comparing both planning approaches, the first one implies quite a lot buffering (100 minutes!). You already know a buffer does not have to be that big, but how to calculate it? Following are two methods giving similar results:

• Goldratt abreviated method: Set the buffer duration as the half the sum of the local safety for all activities. That is: (5+30+5+30+30)/2 = 50 minutes. So you should leave 2 hours before.

• Mike Cohn method: Set the buffer duration as the square root of the sum of squares of the local safeties of all activities (since 90% percentil minus 50% percentil aprox. equals 2σ). That is: (5^2+30^2+5^2+30^2+30^2)^0,5 = 52 minutes.  So you should leave 2 hours and 2 minutes before.

Click here to read the Spanish version for this article

Click the label English to read other English written articles